West Midlands Fire Service Animal Rescue Incident Data

The aim of the exercise is to explore how you approach undertaking analysis on a novel data set and the insights that you can draw from data.

Jurgen Mitsch

2023-07-14

Incident Data - Overview

Data Characteristics

• Date range: Apr 2013 - Mar 2023
• 1411 Total animal rescue incidents
• 7 Districts, 190 Wards in the West Midlands

Key Observations

• Annual total nearly doubled over time
  13/14 (99) to 22/23 (185)
• Notable increases in 14/15, 16/17, and 20/21

• Incidents are infrequent day to day
• Frequency of days with incidents has increased
  (24% in 13/14 to 40% in 22/23)

• Weak annual seasonality

• statement on time between

• Incidents over time
• Trends

Figure 1: Total animal rescue incidents by financial year (top) and by week (bottom).

Figure 2: Total animal rescue incidents by month year-on-year (top) and annual trend component based on fitted prophet model (bottom)

Animal Types

Key observations

• Cat and dog incidents are the majority (64%)

• Cat incidents:

  • account for 57% of total increase
    (since 13/14)
  • have increased by 37.5%
    (since 19/20)

• Dog-related incidents dropped in 20/21 but have since returned to 19/20 levels

• Data on pet population over time somewhat ambiguous

• Ca. 20% of incident descriptions mention RSPCA prior to 17/18 but then drop to 0
• Analysing incident descriptions:
  • Cats getting stuck on trees or roofs
  • Dogs getting stuck in gates and fences
  • Birds getting stuck in fishing gear

incidents by animal type over time

Figure 3: West Midlands Fire Service animal rescue incidents: Total incidents broken down by year and aninaml type. Total incidents by financial year (left) and overall breakdown across all years (right) shown.

Incident Location

Key observations

• Birmingham
  • accounts for ca. 38% of all incidents (2013-2023)
  • contributes ca. 47% of incident increase since 2013
• Highest incident ratios per 10,000 people:(since 20/21, population from 2021 census)
  • Dudley (0.78)
  • Walsall (0.7)

• Data very sparse at Ward level, difficult to interpret

• District level
• Ward level

Make this Notebook Trusted to load map: File -> Trust Notebook

Figure 4: District level visualisation of incident and population geography.

Make this Notebook Trusted to load map: File -> Trust Notebook

Figure 5: Ward level visualisation of incident and population geography.

Where next?

What’s the question?

Stakeholder input is crucial for:

• defining the ask
• interpretation of data and context
• avoiding ‘number blindness’
• steering, prioritising, and reviewing progress (AGILE)

Example questions:

• What factors might impact observed trends?
• Empirical ‘thresholds of significance’ for incident count or frequency?
• Are Fire Services the only entity responding to animal rescue incidents?
• Which fire service team(s) can respond?
• Data quality considerations?

• More data? - depending on the question!
  • ‘Unresponded’ incidents or ‘false alarms’?
  • ‘Call source’ and response times?
  • What happens to the animals afterwards?
  • Fire station / response team locations?

Parallels to Healthcare

• Ambulance Services

• Interaction of system-level services and organisaitons
  • Discharge pathways and destinations
  • Admission routes to healthcare providers
  • Modelling of similarly sparse services/treatment pathways

• Health Inequalities
  • Access to healthcare services
  • Geographic / demographic split
  • Prevalence of conditions/comorbidities

Appendix

Year on year change of incident totals

Figure 6: Year on year change in total incidents by financial year (Mar-Apr).

Total incidents by month

Figure 7: Total animal rescue incidents by month (blue) and 12 month rolling mean (red dashed).

Incidents by weekday

Figure 8: West Midlands Fire Service animal rescue incidents broken down by weekday

Time between incidents

Figure 9: Time between incidents in days

Ratio of days in the (financial) year with incidents

Figure 10: Ratio [%] of days in each financial year where animal rescue incidents have occured.

Incidents by district over time and animal type

Figure 11: Total Incidents by financial year by district (left) and Birmingham incidents by animal type (right).

Animal Populations over time

Figure 12: Cat, dog, and rabbit population over time in millions - Source: PAWS